Integrated Quality of Service for Offline and Online Services in Edge Networks via Task Offloading and Service...

Area of Science:

• Edge computing
• Distributed systems
• Artificial intelligence

Background:

• Edge servers manage both online and offline digital twin (DT) services, impacting resource availability.
• Existing research often neglects the resource demands of offline DT services, hindering online task processing efficiency.

Purpose of the Study:

• To investigate the integrated quality of service (QoS) for both online and offline edge services.
• To develop an optimization strategy for service caching and task offloading considering resource constraints.

Main Methods:

• Formulated a utility maximization problem to balance online service throughput and offline service deadlines.
• Addressed the NP-hard problem by reframing it as a Markov decision process (MDP).
• Introduced a joint optimization algorithm leveraging the deep Q-network (DQN).

Main Results:

• The proposed algorithm effectively considers resource usage from both online and offline services.
• Experimental results show a significant enhancement in overall QoS utility.
• The algorithm improved utility by at least 14.01% compared to baseline methods.

Conclusions:

• The joint optimization of service caching and task offloading is crucial for efficient edge computing.
• The DQN-based approach provides an effective solution for complex resource allocation problems in edge environments.
• This work offers a practical method to enhance the performance of edge servers hosting diverse digital twin services.